[00:00:13] bearND: i'll try to [00:00:25] niedzielski: So, I assume you had the device hitting the MW API [00:00:50] bearND: yes [00:11:39] dbrant mdholloway bearND: nevermind, i just reproed on the API AOSP emulator :( i'm calling off the release. ticket is T130086 [00:11:39] T130086: [Bug] History is garbled - https://phabricator.wikimedia.org/T130086 [09:38:01] morning [09:43:05] phuedx: they've added the next/prev for commits when reviewing! https://github.com/blog/2123-more-code-review-tools [09:44:28] nice [09:50:03] inorite!? [09:50:06] morning btw [09:50:17] joakino: i can't wait to use 'em :D [09:52:59] yeah! [10:35:42] now down to 20 emails! [10:35:43] BOOm [12:43:14] * hashar send a few hundred emails to phuedx [12:43:34] got 789 unread ones :( [12:56:54] what [12:56:55] the [12:56:58] :/ [12:57:01] 789 [13:23:37] +5 --> 794 now [13:23:58] 256 from Gerrit [13:24:04] 260 from Phabricator [13:24:12] 213 in the main inbox, mostly stuff I want to remember about / todo [13:26:06] ^ lol [13:26:36] and I have stopped reading my personal emails, instructed relatives to call me instead [13:26:44] :/ [13:26:53] i'm either doing some right [13:26:57] or i'm incredibly insular [13:26:59] which is wayyyy more effective ;-} [13:27:00] or both! [13:27:27] I just happen to be watching high traffic repositories [13:27:35] and high traffic projects [13:27:41] + subscribed to a large number of tasks :-( [13:27:49] releng is hard [13:28:28] I think about it as something like: lot of devs > ---- releng -----< ops [13:30:23] ^ that [13:30:27] that a million times [13:30:43] then we are only 6 + 1 and have our own internal projects :D [17:01:49] joakino: ping [17:01:51] joakino: pinnnnng [17:17:07] bearND: kaity standup [17:17:16] oooor kaity_ [17:17:24] mbinder: am i in the wrong hangout? nobody there [17:17:30] yup [17:17:34] https://plus.google.com/hangouts/_/wikimedia.org/combined-apps?authuser=0 [17:18:17] mbinder: my invite still had android in the link [17:18:26] dbrant: niedzielski mbinder bearND going to exit and brb [17:18:57] bearND: hrmm, might be because of the recent calendar changes. Do you see "Combined Apps Standup" on your cal? [17:19:35] mbinder: no, it still says Android standup [17:21:52] bearND: refreshed lately? [17:28:59] mbinder: just did. still the same [17:29:21] mbinder: I'm going to delete it and let you send it again [17:29:38] ok [17:30:13] mbinder: oh, wait. Nvm, got it now [17:30:53] mhurd_afk: is ios really blocked on mw-mobilefrontend-leadsection ? [17:31:19] no [17:31:33] dbrant: what do you think should be the starting percentage for the prod app MCS rollout? [17:32:17] bearND: probably no more than 5% [17:33:24] niedzielski: it was just something which degraded the layout and our fix isn't live yet u(ntil 5.0.1 goes live) [17:34:00] mhurd_afk: hm, do you want me to say it's a blocker in scrum of scrums? [17:34:28] niedzielski: ya can u update the ether pad? I'm afk atm [17:35:33] dbrant: ok, let's do 2% then. I'm leaning towards Thursday morning SWAT deploy BTW. //cc:niedzielski mdholloway [17:35:42] of next week [17:37:26] bearND: sounds good to me [17:38:33] mdholloway: this is mainly to give us time to get the lead section stuff back in [17:43:11] dbrant: mdholloway gwicke : updated schedule of MCS rollout https://phabricator.wikimedia.org/T126934 (gwicke feel free to add your project for tracking purposes) Let me know if you'd like this to change. [17:47:23] dbrant: did that auth api spike not make it into the sprint? [17:49:31] niedzielski: ah shoot, forgot to create that task. Here it is (in the sprint): https://phabricator.wikimedia.org/T130137 [17:49:36] niedzielski: thx for reminding [17:50:31] dbrant /cc bearND mdholloway: cool. that would be a good next task for anyone looking for work [18:11:29] dbrant bearND mdholloway: are we waiting on anything else to merge up or should i start on the beta? [18:11:46] hey mbinder sorry forgot to let you know i was in a workshop this morning [18:11:59] niedzielski: good 2 go from me [18:13:42] niedzielski: all systems go [18:14:50] niedzielski: go for it [18:19:30] bearND: added an update [18:21:25] gwicke: sounds great. Thank you! [18:39:52] jdlrobson: amir was the community mentor for the project, not wmf's amir [18:40:07] hey joakino [18:40:20] ahhh what's amir's surname? [18:40:27] or wiki name [18:48:47] nzr: around for some sign off? [18:55:14] nzr: Please preview the new language selector when you have time http://reading-web-staging.wmflabs.org/wiki/Selenium_language_test_page?mobileaction=beta [18:55:46] nzr: http://reading-web-staging.wmflabs.org/wiki/Selenium_language_test_page?useformat=mobile&mobileaction=beta /cc jdlrobson [18:56:04] heh [18:56:14] ;) [18:56:17] just in case [18:56:17] jdlrobson: he's the creator of the community wishlist entry https://fa.wikipedia.org/wiki/%D8%A8%D8%AD%D8%AB_%DA%A9%D8%A7%D8%B1%D8%A8%D8%B1:4nn1l2?uselang=en [18:56:38] joakino: sweet [18:56:38] don't mind me, just provision mwv /again/ [18:56:41] sounds like a good person to have :) [18:56:56] yep [19:05:48] jdlrobson: able to make the product planning meeting? [19:06:20] jdlrobson: if not you, bmansurov or jhobs could one of you attend? ^ [19:06:41] dr0ptp4kt: what is this now? [19:06:42] im in sorry [19:10:05] bmansurov: any reason you're not making the LESS change? I've recommended it twice now and heard no comment back from you yet. [19:10:49] bmansurov: to be clear, I'm referring to this: https://gerrit.wikimedia.org/r/#/c/275881/18..20/resources/skins.minerva.beta.styles/pageactions.less [19:13:12] jhobs: let me check [19:13:25] dr0ptp4kt: what time is it? I can attend, can get an invite? [19:13:38] bmansurov: disregard, we're good [19:13:47] ok [19:14:08] jhobs: i made the change, looks like i forgot to add it to git [19:14:22] bmansurov: ah ok, cool. +1 from me then on the next PSA [19:14:24] PS* [19:15:51] jhobs: check now [19:16:23] jhobs: sorry for missing your comment earlier [19:19:25] bmansurov: that wasn't quite the right fix. The #language-switcher rule should still be within #page-actions, just not within the li. Where it is now isn't specific enough to override the float:right. I'll comment on the task for visibility as well. [19:19:31] and no worries about the comment [19:20:02] why does it have to be inside page actions? [19:20:05] jhobs: ^ [19:20:16] oh [19:20:17] wait [19:20:24] bmansurov: https://gyazo.com/ba9ac6d4bd3fffecd65976f04ae12198 [19:20:55] then we should loosen the other rule [19:21:05] no need to have nested IDs imo [19:21:29] bearND: here we go. this guy is returning "true" for revision: https://test.wikipedia.org/w/api.php?action=mobileview&format=json&formatversion=2&prop=text%7Csections%7Clanguagecount%7Cthumb%7Cimage%7Cid%7Crevision%7Cdescription%7Clastmodified%7Cnormalizedtitle%7Cdisplaytitle%7Cprotection%7Ceditable&onlyrequestedsections=1§ions=0§ionprop=to [19:21:29] clevel%7Cline%7Canchor&noheadings=true&page=Foo_%28disambiguation%29&thumbsize=640 [19:23:02] bmansurov: i'm not sure how we would loosen the other rule. I think it's fine to have the nested IDs in this case since we want the language switcher to appear on the left but it can't be inserted into the DOM before the other two [19:23:11] bearND: https://phabricator.wikimedia.org/P2777 [19:23:21] bmansurov: (at least I assume that's why we're floating it left) [19:24:35] jhobs: how about we float each icon specifically? [19:24:36] niedzielski: hmm, a bug in formatversion 2? [19:25:29] bearND: without formatversion=2, revision is an empty string [19:25:55] bmansurov: I suppose we could do that, yeah. It would start to get a bit more complicated if we added even more page actions, but with only 3 atm, that seems reasonable [19:26:09] so just a float:right on the watchstar and float:left on the language switcher then? [19:26:31] and float right on the pencil [19:26:58] jhobs: it will get complicated either way because we have to add specific rules if we want to float the new action to the left [19:27:35] bmansurov: true. And we shouldn't need to float the pencil right I don't think. [19:27:40] bearND: maybe this is only an issue on the foo disambiguation page of test wikipedia [19:27:54] bmansurov: one on the left, one on the right == remainder in the center [19:28:47] jhobs: you can't have it in the center, by default inline-blocks are floated left [19:28:56] jhobs: so we could just float the pencil and watchstar right only [19:29:26] dbrant: you need me for grooming? [19:29:36] bmansurov: I don't see a style in production floating it left. [19:29:43] (observing web planning) [19:30:02] mbinder: nope, you can skip if you like [19:30:07] dbrant: k [19:30:10] jhobs: i meant stacked to the left by default [19:30:14] jhobs: no need to float [19:30:39] bmansurov: right, but float will override that so you shouldn't need to float it right. 3 less lines of less. Lemme double-check my thinking in a sandbox real quick [19:32:14] bearND: mtg? [19:33:56] dbrant: omw [19:34:01] bmansurov: I just achieved the following image by floating the notification icon left and the watchstar right without floating the pencil. The order they appear in the DOM is pencil, watchstar, notification. https://gyazo.com/5e269de8f63540541a4a17234280e98a [19:34:17] (basically I used the notification icon as the language switcher since I'm just editing the production site) [19:34:53] jhobs: try making the window smaller [19:35:01] do you see the notification icon to the left? [19:35:31] the pencil floats left too right? [19:36:19] bmansurov: ah there's the culprit of confusion! Ok cool so you're right we'll need the three floats. Incidentally, what actually happened was the two floated elements broke onto their own line, so it was even worse. [19:36:44] ok, let me push a patch real quick [19:37:04] jhobs: pushed [19:38:08] bmansurov: you got rid of the float:left for the language icon. We'll still need that or it'll break onto its own line I think [19:38:27] try it [19:38:45] we have display: inline-block above [19:38:53] bmansurov: yeah I should probably do that before commenting huh? :D [19:39:03] good joke [19:42:00] bmansurov: oh I missed a design change apparently anyways, that's why I was so confused. Alright looks good to me [19:42:24] waiting for @nzr to sign off now [19:57:03] bmansurov: is the structured overlay to stable patch supposed to depend on the new beta icon? I thought the two were being done in parallel? [19:57:24] jhobs: no, but i'd rather not rebase [19:57:54] bmansurov: well it needs a rebase regardless because of the new PS, but I guess that's fine since the other one is just waiting on design review [19:58:16] if it gets held-up by the icon at all, we should rebase it though. [19:58:34] niedzielski: bearND: mdholloway: right, so the issue is that "revision" is an empty string... [19:58:58] dbrant bearND dbrant: or a boolean [19:59:01] jhobs: ok [19:59:43] niedzielski: formatversion=2 turns empty string into boolean true. So, what dbrant said [20:00:26] niedzielski: bearND: mdholloway: so how can we update our pojo Mobileview class to ignore the revision field, if it's malformed? [20:02:46] dbrant: if we didn't use formatversion=2 then it would be automatically ignored. Otherwise we would need to come up with a Gson hack since the type is obviously wrong. [20:03:55] on potential hack could be to treat the field as a string and try to convert it to a long. But it's def. a hack [20:04:08] one* [20:04:34] dbrant bearND dbrant: maybe this has something to do with it https://git.wikimedia.org/commit/mediawiki%2Fextensions%2FMobileFrontend/82f2fe37256ca7a01d537aa449aaf16857f5ac8a [20:06:16] dr0ptp4kt: comin'? [20:06:47] jdlrobson: hello! we're investigating an issue in the android app. we think the mobileview page revision should return a number but we get a boolean on formatversion=2 and an empty string otherwise. is this expected? e.g. https://test.wikipedia.org/w/api.php?action=mobileview&format=json&prop=text%7Csections%7Clanguagecount%7Cthumb%7Cimage%7Cid%7Cr [20:06:48] evision%7Cdescription%7Clastmodified%7Cnormalizedtitle%7Cdisplaytitle%7Cprotection%7Ceditable&onlyrequestedsections=1§ions=0§ionprop=toclevel%7Cline%7Canchor&noheadings=true&page=Foo_%28disambiguation%29&thumbsize=640%20596 [20:07:36] niedzielski: bearND: we actually don't use the revision id for anything significant... [20:10:01] dbrant: looks like only for EL. If you don't need that we could drop it completely. [20:16:40] bearND: don't we use it in the content service to avoid differences between page revisions over multiple requests? [20:20:28] niedzielski: bearND: mdholloway: then how about we go with bearND's "hacky" idea of treating revision as a string? It's not *that* bad... [20:22:22] niedzielski: bearND: dbrant: i'd be keener on not using formatversion=2 [20:22:35] niedzielski: bearND: dbrant: iirc the benefits are minimal [20:23:25] mdholloway: we are using formatversion=2 [20:23:42] niedzielski: right. i mean ceasing to use formatversion=2 [20:24:45] mdholloway: sorry, i'm completely missing the idea :) formatversion=2 returns a boolean, removing it returns an empty string. in either case, we have to change the Gson data type [20:24:55] mdholloway: i think the reason for keeping formatversion=2 is for consistency sake, and there are some boolean fields we benefit from when parsing via Gson [20:25:48] niedzielski: i was just going off bearND's statement above that it would be automatically ignored if we didn't use formatversion=2. but yeah, we've added a lot more Gson parsing since I've really thought about it [20:26:57] mdholloway: it's an option. I was just playing devil's advocate [20:27:08] niedzielski: dbrant: bearND: i'm fine with the "hack" [20:27:21] mdholloway: ah, missed that. i don't see how that could work either [20:27:39] mdholloway: the field is present in both formats [20:28:31] dbrant bearND dbrant: does hack = Mobileview.revision becomes a boolean the accesors convert it it to a long? we should probably try to do something a bit sturdier in case it's fixed on the back end [20:29:46] dbrant bearND dbrant: maybe this is only an issue on test wikipedia? see https://en.wikipedia.org/w/api.php?action=mobileview&format=json&formatversion=2&prop=text%7Csections%7Clanguagecount%7Cthumb%7Cimage%7Cid%7Crevision%7Cdescription%7Clastmodified%7Cnormalizedtitle%7Cdisplaytitle%7Cprotection%7Ceditable&onlyrequestedsections=1§ions=0§io [20:29:46] nprop=toclevel%7Cline%7Canchor&noheadings=true&page=Foo_%28disambiguation%29&thumbsize=640 [20:29:53] (just changed test to en) [20:30:16] niedzielski: yes, i was just going to say that, too [20:30:56] niedzielski: sorry just catching up [20:31:40] formatversion=2 gives you a nicer API response [20:32:01] for example pages => [ page, page2 ] rather than pages = { id => page, id2 => page2 } [20:32:05] amongst other things [20:32:11] niedzielski: dbrant : mdholloway : looks like this is only an issue when using the MW API, not RB. So, we'd only have to apply the hack to MwPageLead [20:32:11] jdlrobson: hey! we're seeing some funky issue on test wikipedia. the revision is a boolean on test but a long everywhere else. the app currently expects the latter [20:32:41] if it's really worth it, considering it's seems to be only on testwiki [20:33:30] we did fiddle with revision id recently so highly possible we've introduced a bug [20:33:35] lemme see what testwiki is running [20:33:50] also can you replicate issue on the beta cluster? (http://en.m.wikipedia.beta.wmflabs.org/w/api.php) [20:35:14] so testwiki is running 1.27.0-wmf.17 (so enwiki will have same issue on thursday) [20:35:17] enwiki is running 1.27.0-wmf.16 niedzielski [20:35:25] (https://en.m.wikipedia.org/wiki/Special:Version) [20:36:35] jdlrobson: same happens on beta [20:36:36] https://en.wikipedia.beta.wmflabs.org/w/api.php?action=mobileview&format=json&formatversion=2&prop=text%7Csections%7Clanguagecount%7Cthumb%7Cimage%7Cid%7Crevision%7Cdescription%7Clastmodified%7Cnormalizedtitle%7Cdisplaytitle%7Cprotection%7Ceditable&onlyrequestedsections=1§ions=0§ionprop=toclevel%7Cline%7Canchor&noheadings=true&page=Barack_Obama&thumb [20:36:36] size=640 [20:36:48] bearND: double confirm :) [20:37:15] niedzielski: bearND you file a bug? [20:37:17] i can look into it [20:37:55] jdlrobson: not yet. we thought it might be some weird app issue for a bit [20:38:08] i'll file one [20:39:36] i can replicate locally [20:39:54] does this mean the app will break on thursday? [20:40:51] dbrant: i think so... i'm seeing different behavior on different devices [20:41:58] niedzielski: what's the effect on the app? The pages still load fine for me [20:43:34] bearND: on one device it loads fine for me, on another i get this https://veuwer.com/i/3piq [20:44:09] niedzielski: ah, thanks. That's very bad [20:46:41] So, in this case it was good we ran tests against testwiki. Gave us a couple of days before it blew up. [20:47:04] quite! [20:50:33] dbrant bearND mdholloway jdlrobson: ticket for tracking is T130155 [20:50:34] T130155: [Bug] Mobileview returns bad revision IDs - https://phabricator.wikimedia.org/T130155 [20:51:26] dbrant bearND mdholloway: so app side, do we want to try to accept both kinds of revisions (booleans and longs)? [20:52:06] bearND: good point about testing against testwiki! [20:54:58] niedzielski: well, if it will be fixed in the API, then it's probably not necessary... [20:55:06] niedzielski: I'd prefer to just have this be fixed on the server side. I expect we'll have a SWAT deploy or halt the train. [21:02:30] dbrant bearND mdholloway: is there anyone we should check in with on that? [21:04:31] proly jdlrobson [21:08:15] bearND: niedzielski i've found the problem let me see if i can swat it [21:08:41] https://gerrit.wikimedia.org/r/277890 < need to find someone to merge that though [21:12:41] jdlrobson: \o/ [21:13:45] hey phuedx didnt realise you were still around :) [21:26:22] nzr: regarding my comment in https://phabricator.wikimedia.org/T126836#2124052 is it sufficient to just collapse the TOC and skip creation of a mock? [21:26:39] nzr: if so, i'll update the commentary on it and we can keep it simple [21:27:18] (or if you want to, please feel free...we have a couple days i think before it percolates to the top of the prioritized backlog that is sprint 69) nzr [23:02:47] dbrant bearND mdholloway: i'll be publishing the beta shortly unless i hear any objections [23:03:52] +1 [23:04:16] +1 [23:19:53] niedzielski: i'm getting the fix deployed [23:20:03] niedzielski: would be great if you could be around to test [23:20:05] jdlrobson: our hero! yayyy! [23:20:12] jdlrobson: sure [23:20:17] jdlrobson: literally the least i could do [23:23:02] niedzielski: i will ping you when it is live [23:23:14] jdlrobson: sounds good [23:50:02] niedzielski: can you verify? [23:50:12] niedzielski: it LGTM https://test.m.wikipedia.org/wiki/Special:ApiSandbox?useformat=desktop#action=mobileview&format=json&page=Barack+Obama&prop=text%7Csections%7Cnormalizedtitle%7Crevision&formatversion=2 [23:51:35] jdlrobson: was just about to ping you. it's working fine for me on two devices. i'll try a couple more but i think it's all good from our end [23:51:54] jdlrobson: thanks so much! [23:52:20] sweet [23:55:15] yeah looks fine. thanks again :)